Big tech has spent two years cutting engineers because of AI, and Google is now testing the opposite idea by letting candidates bring AI into the interview itself.
It is one of the clearest signals yet that the bar for software engineers is shifting away from solo coding and toward working alongside an AI assistant.
Google is rolling out a new interview process for software engineering jobs that allows the use of an "approved" AI assistant during the coding round, according to an internal document reviewed by Business Insider. The format applies to junior and mid-level roles, starting with select teams in the US and scaling across the company if it works.
Beginning in the second half of the year, candidates in the new "code comprehension" round will be expected to read, debug, and improve an existing code base with help from the AI - and interviewers will judge "AI fluency, including prompt engineering, output validation, and debugging skills."
Brian Ong, Google's vice president of recruiting, said the company is updating its process "to be more reflective of how our teams are operating in the AI era," with the first orgs to pilot the format including Cloud and Google's platforms and devices unit.
Three out of every four lines of new code inside Google are now written by AI, the company said in April, which is the current state of how Google ships software rather than a forecast.
OpenAI president Greg Brockman recently said AI has moved from writing 20% of code to 80% across the industry, which makes a no-AI coding interview look increasingly out of step. Emily Cohen, who runs people and operations at AI coding startup Cognition, told Business Insider that blocking AI in interviews is like asking a kid to take a math test without a calculator.
Other companies got there first, with Canva and Cognition already letting candidates use AI in technical interviews. Google moving in the same direction matters because it sets the standard for how the rest of big tech hires.
The Google document also outlines other changes. The "Googleyness and Leadership" round, normally focused on behavior, will now include a technical design discussion about a candidate's past project, while junior candidates will get a new round focused on open-ended engineering challenges.
Together, the changes show what kind of engineer Google now wants - less rote coding, and more design thinking, debugging, and prompt skill, the things that hold up in a workflow where AI does the typing.
For investors, the read is straightforward: Google is signaling that its productivity per engineer is going up, which is consistent with management's broader message that AI is changing both the cost base and the size of teams it needs to ship product.
The internal document calls the new format "human-led, AI-assisted," and that phrase is the whole story.
If Google rolls this out across the company, expect every other major tech employer to follow within 12 months. The new test is whether you can run AI well, not whether you can outrun it.